The infrastructure question is no longer only where capacity can be built

A data center project can look viable on paper and still face major power challenges. That makes data center carbon emissions an infrastructure planning issue. Electricity demand is rising as operators add more computing capacity. The International Energy Agency says data center electricity use increased about 17% in 2025. Global electricity demand grew about 3% during the same year. Electricity systems are also managing new industrial loads and wider electrification. Many regions face grid connection and transmission constraints. New facilities can require additional grid, generation and transmission infrastructure. The exact requirements depend on the location and project design. Carbon intensity adds another layer to this decision. Electricity mixes differ across power systems, so data center emissions can vary by location. This changes the infrastructure conversation. Operators cannot look only at available power capacity. They also need to understand where that power comes from. The location of a facility can affect its long-term energy profile. That makes carbon performance increasingly relevant to site planning.

Should carbon intensity influence where the next data center gets built?

Data center location decisions involve several practical factors. These can include power, connectivity, land and water availability. Network access and customer proximity can also influence site selection. Each project has different priorities and constraints. Carbon intensity adds another variable to that decision. Two locations can have similar power access but different electricity mixes. Their indirect emissions can therefore differ. This distinction matters for facilities that consume large amounts of electricity. The IEA estimates that data centers produce about 180 million metric tons of indirect CO₂ emissions from electricity use. That figure excludes emissions from backup power generation. The IEA also expects data center emissions to increase significantly through the decade. That raises an important planning question. Should a location be selected only because it has enough power? Or should its electricity mix also influence the decision? For infrastructure planners, the answer is becoming harder to ignore.

A megawatt is no longer enough information for a site decision

Available power remains critical for every data center project. However, megawatts alone do not explain the full infrastructure picture. Grid capacity, connection timing and local congestion also matter. These factors can determine how quickly a facility can become operational. The IEA says more than 2,500 GW of generation, storage and large-load projects remain in global grid connection queues. Data centers are part of this growing demand for grid connections. A site with available capacity can still require grid upgrades. Additional generation or transmission may also become necessary. Transmission projects can take years to complete. The IEA says new transmission can take four to eight years in advanced economies. Transformer and cable supply constraints can add further pressure. For end users, these delays can have practical consequences. New computing capacity may take longer to reach the market. Regional infrastructure choices may also become more limited. The planning question is therefore bigger than a simple megawatt calculation.

Can renewable procurement solve a problem created by physical infrastructure?

Renewable energy procurement has become an important part of corporate energy strategies. It can help organizations address emissions linked to electricity consumption. However, renewable procurement does not remove the physical power needs of a data center. A facility still needs reliable electricity at the site. A renewable energy contract does not replace the required grid connection. It also does not remove local transmission constraints. Physical electricity supply and contractual energy procurement address different parts of the problem. Grid queues can delay projects even when operators have strong sustainability goals. Transformer shortages can create additional delays. Transmission requirements can also affect development schedules. The IEA identifies these infrastructure constraints as important barriers to faster data center expansion. Operators may therefore need a wider planning approach. That approach can include grid conditions, energy contracts and storage. It can also consider onsite generation and demand flexibility. Renewable procurement remains important, but it cannot replace physical infrastructure planning.

Is efficiency enough when demand keeps accelerating?

Energy efficiency remains one of the industry’s most important tools. More efficient hardware can reduce energy use for individual computing tasks. Better cooling can also reduce facility-level energy demand. Yet efficiency does not guarantee lower total electricity consumption. The reason is simple. Computing demand continues to grow. AI workloads are adding new requirements for accelerated computing capacity. The IEA says electricity use per AI task is falling rapidly. Overall data center electricity demand is still increasing. The latest IEA outlook expects global data center electricity use to reach about 945 TWh by 2030. That would be roughly double the 2024 level. AI-focused accelerated servers will account for a large share of the increase. This creates a difficult planning question. Should operators measure efficiency only at the equipment level? Or should they also measure total facility demand? A more efficient data center can still consume more electricity as workloads expand. Infrastructure planning needs to account for both trends.

Cooling has become part of the carbon equation

Cooling is becoming more important as computing density increases. Higher-density systems generate more heat. That creates greater demands on thermal management. The IEA estimates that data center cooling consumed about 80 TWh globally in 2025. Liquid cooling is gaining attention for high-density computing. It can support thermal management in systems with demanding heat loads. Its overall energy performance depends on system design and operating conditions. The electricity used by the cooling infrastructure also affects its environmental impact. Cooling decisions can also affect water use and mechanical infrastructure. Climate conditions can change the requirements of a cooling system. Facility design can influence its efficiency and resilience. These factors make cooling part of wider infrastructure planning. For operators, the decision is no longer simply about removing heat. It is about managing heat efficiently while controlling power demand. That connects cooling directly with the wider carbon and energy discussion.

What happens when the grid becomes the limiting infrastructure?

The biggest constraint may sit outside the data center itself. Grid capacity can determine how quickly new facilities can connect. This issue is becoming more important as data center demand grows. The United States provides a useful example. The U.S. Department of Energy reported that data centers consumed about 176 TWh in 2023. That represented 4.4% of total U.S. electricity consumption. DOE-backed research projected data center electricity use could reach 325 TWh to 580 TWh by 2028. The IEA has also highlighted grid connection challenges. It estimates that grid constraints could delay around 20% of planned global data center capacity through 2030. Projects can therefore face delays while waiting for grid connections and infrastructure upgrades. These constraints can influence where developers build. Locations with faster access to power may become more attractive. This could affect regional development patterns. It could also increase pressure on electricity systems around major data center clusters.

The next planning model needs to start with the end user

The carbon debate becomes more practical when infrastructure affects customers. Enterprises depend on cloud platforms and AI services every day. They need reliable capacity and predictable performance. They also need infrastructure that can scale with demand. Power constraints can affect how quickly new capacity becomes available. They can also influence where operators build new facilities. The IEA says many U.S. data centers under development are concentrated in existing clusters. Such concentration can increase pressure on local electricity systems. Electricity is also a major operating input for data centers. Changes in power costs can therefore affect infrastructure economics. Customers may eventually feel those pressures through service costs or deployment decisions. The effect will vary by market and business model. Environmental reporting adds another consideration. Enterprises may want greater visibility into the energy supporting their workloads. They may also need better information about associated emissions. That makes infrastructure transparency increasingly relevant to end users.

The harder question is whether infrastructure planning will change fast enough

The data center industry is entering a period of much higher electricity demand. The latest IEA data shows that trend clearly. Data center electricity consumption increased about 17% in 2025. That added around 70 TWh of global electricity demand. The IEA expects data center electricity consumption to reach about 945 TWh by 2030. AI-related accelerated servers will drive a large share of that growth. At the same time, electricity systems must support industry and transport. They must also meet growing cooling and electrification demand. That creates a much bigger infrastructure challenge. The issue is not simply how many data centers the industry can build. The issue is whether the surrounding power system can support them. Grid connections, transmission and generation must expand alongside digital infrastructure. Carbon emissions add another dimension to that challenge. The electricity source can influence the emissions associated with data center operations. Site selection can therefore affect more than construction schedules and power availability. The industry now faces a harder question. Can data center growth align with reliable power and lower-emissions electricity? The answer will depend on decisions made before facilities are built. It will also depend on how quickly energy infrastructure can respond. For end users, the outcome matters most. They need digital infrastructure that can scale without sacrificing reliability. They also need predictable costs and performance. The next generation of data center planning will need to address all three. Data center carbon emissions are therefore becoming more than a sustainability discussion. They are becoming part of the infrastructure decision itself. That shift could change how operators evaluate power, locations and expansion plans. It could also determine how effectively the industry meets growing demand.